This invention provides a combustion apparatus which can be used in a high-temperature environment, such as in a gas turbine.
Gas turbines have been used in the generation of electric power. Considerations of efficiency require that the turbines operate at high temperatures, for example, 2300.degree. F. or greater. In a conventional thermal (i.e. non-catalytic) combustion apparatus, all of the fuel is mixed with part of the combustion air, and the mixture is burned in a flame. The flame temperature exceeds 3000.degree. F., and at such high temperatures, nitrogen oxides (NO.sub.x) are formed from the nitrogen in the air. The combustion gas at 3000.degree. F. or higher is mixed with the remainder of the combustion air to cool the mixture to about 2300.degree. F., which is a safe temperature for injecting into the gas turbine. The production of unacceptably large amounts of nitrogen oxides is a major disadvantage of conventional gas turbine systems.
To eliminate the production of NO.sub.x, it has been proposed to burn the fuel at a lower temperature and to complete the combustion with a catalyst. By adjusting the amount of fuel so as to keep the temperature of combustion below about 2300.degree. F., one can eliminate the production of NO.sub.x. However, at the lower temperatures, the combustion is not complete, resulting in an emission of unburned hydrocarbons. Furthermore, natural gas and air may not be uniformly mixed, thus permitting regions of combustion exceeding 3000.degree. F. A catalytic converter assures mixing and produces a homogeneous flame front under 3000.degree. F.
However, at temperatures of 2300.degree. F. or greater, metal catalysts such as platinum or palladium cannot be used. At such temperatures, the metals are volatile. For this reason, the useful lives of catalytic converters of the prior art, when used in high-temperature environments, are unacceptably short.
It has been known to use oxides of rare earth metals as catalysts in high-temperature environments, such as in the operation of gas turbines. For example, terbium oxide, applied to a ceramic support, has been used. Terbium oxide has sufficient activity at 2300.degree. F. to complete the combustion, and will withstand the high temperatures of the gas turbine. Another class of materials believed to be practical as high-temperature catalysts are the refractory metal silicides described in copending U.S. patent application Ser. No. 279,455, filed Dec. 5, 1988, entitled "A Gas Modifying Reactor and Method of Modifying a Reactive Gas Composition".
The ceramic supports used in the prior technology have major disadvantages. Ceramic supports are likely to shatter due to thermal shock.
If one wants to provide a practical combustion apparatus, it is necessary to build the apparatus in stages. An ignition stage, having a metal catalyst, is used to start the combustion. The subsequent stages, operating at high temperatures, carry the combustion to completion.
Even in a two-stage apparatus, there is still the danger of thermal damage to the metal catalyst. It has been suggested that the temperature in the ignition stage could be controlled by simply reducing the length of the ignition catalyst so that the first stage of combustion is less complete. But it has been shown both theoretically and experimentally that the ignition catalyst cannot be saved from overheating simply by shortening the combustion channels.
When a fuel-air mixture flows through the channels of a catalytic reactor, the temperature of the mixture increases smoothly as the combustion proceeds. But the temperature of the catalyst-bearing wall of the channel does not increase smoothly. Instead, this temperature increases rapidly near the entrance to the channel and approaches the adiabatic combustion temperature. Then it remains close to the adiabatic combustion temperature over the length of the channel.
If the channel is too short to achieve complete combustion, the temperature of the exiting gas will be below the adiabatic combustion temperature. Even so, the temperature of the catalyst-bearing wall will be close to adiabatic, over all but the front end of the channel.
This behavior of the temperature at the channel wall is not anomalous. It is predicted by mathematical models of the combustion process. Recent experimental work confirms the models. The work was done by D. A. Santavicca at Pennsylvania State University under contract from the U.S. Air Force Office of Scientific Research, Contract No. AFOSR84-0224. He constructed a stack of catalyst-bearing metal plates which were 50 mm wide and 100 mm long, and having a thickness of 1 mm. The plates were spaced 6 mm apart. The fuel-air mixture flowed parallel to the 100 mm length of the plates. Tiny holes were drilled into the edges of the plates at distances, from the leading edge, of 3 mm, 38 mm, and 97 mm. Thermocouples in the holes showed that the temperature of the catalyst-bearing surface does indeed rise to the adiabatic temperature, close to the leading edge, and remains at the adiabatic temperature over the remainder of the flow path.
Thus, even if the catalytic combustion channel is shortened, the catalyst-bearing walls will still reach the adiabatic temperature and the catalyst will be deactivated. It is therefore necessary to cool these walls with a part of the fuel-air mixture that is not being combusted. The present invention provides a structure for cooling the walls. This structure can be used with a gas turbine, such as in the environment described in the above-identified copending patent application, and in other environments.